JOURNAL OF LANGUAGE AND LINGUISTIC STUDIES - Journal of Language and Linguistic ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Available online at www.jlls.org
JOURNAL OF LANGUAGE
AND LINGUISTIC STUDIES
ISSN: 1305-578X
Journal of Language and Linguistic Studies, 17(2), 706-719; 2021
The English Google translation of Indonesian lecturer’s academic writing: A
preliminary study
Menik Winiharti a 1 , Syihabuddin b , Dadang Sudana c
a,b,c
Indonesia University of Education, Bandung, Indonesia
a
Bina Nusantara University, Jakarta, Indonesia
APA Citation:
Winiharti, M., Syihabuddin, S. & Sudana, D. (2021). The English Google translation of Indonesian lecturer’s academic writing: A
preliminary study. Journal of Language and Linguistic Studies, 17(2), 706-719. Doi: 10.52462/jlls.49
Submission Date: 18/02/2021
Acceptance Date: 02/05/2021
Abstract
The work of translation seems to be much easier with the assistance of web-based Machine Translation such as
Google Translate. Does it work well? This study aims at finding how Google translates academic texts from
Indonesian into English. It is also to find the extent to which Google Translate accurately and naturally transfers
such texts. The data are Indonesian academic texts written by undergraduate lecturers from three different
majors: Management, Japanese Literature, and Mathematics. The discussion section in each article is translated
into English using Google Translate web. The data is analyzed in terms of two perspectives: forms and meaning.
With regard to form or syntactic analysis, the data are investigated based on the structure of the Source Language
and the Target Language. It is also to observe whether such forms are natural to generate meaning in the Target
Language. With regard to the meaning or semantic analysis, the data are evaluated based on the original
message, whether or not the message conveyed in the Target Language is accurate. The findings indicate that
most of the translations of Google Translate are built in similar forms to those of the Source Language. In terms
of meaning, the messages conveyed seem to be fairly accurate even though inaccuracy is still found.
Keywords: Translation; Google Translate; Academic texts; Syntax; Semantics
1. Introduction
The occurrence of machine translation is very much likely to be pioneered in the late-1940s and
begun in the early-1950s when “tentative ideas for using the newly invented computers for translating
natural languages” were discussed. (Hutchins 2006) Years later, begun in the mid-1990, machine
translation has developed quite rapidly, fueled by the internet which provides internet-based online
translation. Nowadays, one of the most popular web-based machine translations is Google Translate.
According to Groves & Mundt (2015), Google Translate (for further use, it is abbreviated as GT) is the
most visible web-based machine translation available today, free of charge and capable of translating
many languages. Similarly, Lotz & Rensburg (2014) stated that GT is a free online application, offered
by Google Inc., which allows users to instantly translate words, sentences, documents, and even
websites. James Kuczmarski, a product manager from GT, stated that this machine has become an
1
Corresponding author.
E-mail address: menikw@upi.edu
© 2021 Cognizance Research Associates - Published by JLLS.707 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
important tool for communicating in cross-languages. (Kuczmarski 2018) However, there has been a
debate toward the success of GT in translating many languages in the world – 103 languages according
to Kuczmarski (2018). On the one hand, GT undeniably has become an important means for people to
communicate across languages, but that does not mean that GT is always perfect in translating all
languages. Khosravizadeh & Pashmforoosh (2011) clearly stated that although GT is an asset that is
always ready to help language users around the world, it also has a number of weaknesses. According
to them, humans are still needed to involve in the editing process to generate a final product of
translation.
1.1. Literature review
Several studies related to the quality of GT translations have been conducted. The most common
problem resulted from GT is mistranslation (Lotz & Rensburg 2014). Furthermore, the product of GT
still shows errors at various levels, varying from word level (Vidhayasai, Keyuravong, & Bunsom
2015), sentence-level (Chen, Acosta, & Barry 2016), to the level of discourse (Groves & Mundt 2015;
Vidhayasai et al. 2015). Problems are also discovered at the semantic level (Allue 2017;
Khosravizadeh & Pashmforoosh 2011); syntax (Allue, 2017; Vidhayasai et al., 2015), and grammar
(Khosravizadeh & Pashmforoosh, 2011; Li et al., 2014). GT also shows different results for different
genres of texts (Allue, 2017; Calude, 2003; Rensburg et al., 2012; Voigt & Jurafsky, 2012). In this
case, a translation of one type of text can show better results than a translation of another type of text.
Yet, GT results are still acceptable concerning the general understanding of information (Li et al.,
2014).
One issue that needs attention is that a number of people are quite dependent on GT for it is easily
accessible and free of charge (only requires an internet network). Vidhayasai et al. (2015) said that the
dependency on translation tools such as GT should become a concern, in that it is to know whether the
machine is efficient and practical. What has happened in the community so far is that Google Translate
has not yet provided a translation result that satisfies its users, especially with regard to its naturalness,
even though the results can be quite useful because it provides instant translation. In this case, they
said that GT has limitations because each language has its own character and linguistic functions.
Their research revealed that the use of GT to translate the 'terms and conditions on an official website
of an airline in Thailand raises several errors at the lexical, syntactic, and discursive levels. Such
imperfect translation results can cause major problems with regard to passenger safety. According to
them, however, GT can be used at the word level but requires a careful assessment from its users in
addition to requiring many revisions. (Vidhayasai et al., 2015)
Furthermore, concerning how GT translates different text genres, Allue (2017) investigated the
accuracy of GT on tourism text and football match reports from two language directions (SL to TL and
vice versa). According to him, GT produced a slightly better translation in the tourism texts than that
in the football match report. In the education field, GT also has a role although limitations are still
found. In other cases, Groves & Mundt (2015) explored the function of GT in student essay writing.
According to them, GT is merely able to translate at the lexicogrammatical level and cannot help
students align their writing to the norms and expectations of the wider discourse. Moreover, when
translating newly coined words, it becomes more difficult for machine translations – including GT – to
generate meaning in the Target Language. Awadh & Shafiull (2020) have compared the performance
of human and machine translators to translate neologism. Their results show that translation students
found it hard to translate neologisms accurately and a machine translation resulted in poorer
translations.
© 2021 Cognizance Research Associates - Published by JLLS.Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719 708
From the point of linguistics view, Erton & Tanbi (2016) have found that students of translations
have a better understanding and achievement in translating and interpreting if linguistics and discourse
courses are embedded in the courses. Thus, linguistics and translations cannot be separated as
linguistics plays a vital role in the field of translation.
In the meantime, GT has been also widely used in Indonesia, including by academicians who are
required to publish scientific articles using the English language. Unfortunately, not all undergraduate
lecturers in Indonesia are well equipped with English writing. In other words, English can be their
obstruction in writing their research reports. (Arsyad et al., 2019; Hartono et al., 2019). With regard to
the use of machine translation for this case, Michaela Panter, an academic editor at aje.com (Panter,
n.d.), states "with the increasing number of research articles by non-native English speakers and a lack
of spare time in researchers' schedules, automated translation may seem like an appealing option."
However, as she further points out, there are four kinds of the most serious errors produced by GT,
namely sentence fragments, long sentences, illogical ordering phrases, and literal instead of context-
dependent translation. In addition, poor sentence construction, syntax, and terminology used can
reduce the readability of a text, which can lead to unclear or even lost meaning. (Panter, n.d.)
Due to those findings, the present study is conducted to find how GT translates the Indonesian
Lecturer’s Academic Writing from Indonesian into English. It is also to find the extent to which GT
accurately and naturally transfers the Indonesian Lecturer’s Academic Writing from Indonesian into
English. As far as I am concerned, the study of translations from Indonesian into English is not yet
much conducted, specifically in academic texts using web-based machine translation.
1.2. Theoretical Framework
Kenny (2011) states that “Empirical translation studies can be divided into two main categories:
those that focus on the translation process, and in particular on the translator’s cognitive activity; and
those that study translation products – target texts which can be related, amongst other things, to their
host cultures, their users, and their respective source texts.” Since the present study is to investigate the
quality of GT, then it focuses on the product, i.e. the translation resulted from a web-based machine
translation – Google Translate.
Basically, translation is transferring the meaning of a source language (SL) into the target language
(TL or Receptor Language: RL). Thus, a work of translation must convey a message and not a form
(Larson 1998). Correspondingly, Baker (1993) revealed about equivalence by asserting that the
translation must be equivalent to the original language, in semantic (meaning) or formal (form)
category.
Furthermore, concerning types of translations, Larson (1998, 17) classifies them into two main
kinds of translations: literal and idiomatic. The first one is the translation which follows the forms of
SL, thus it is form-based. The latter tries to communicate the SL's meaning into the natural form of the
TL, thus it is meaning-based.
With regard to strategies that a translator may apply at the text level, Baker (2018, 179-183)
proposed voice change, change of verb, nominalization, and extraposition. These strategies are
suggested to resolve the debate between syntactic and communicative functions in translation and
language learning. Meanwhile, in terms of the evaluation of translation, Larson (1998, 529-530)
argued that when a translation is done, it should be evaluated based on three factors: accuracy,
clearness, and naturalness. A translation is said to be accurate when it conveys the original message of
the source language. It is said to be clear when the audience of the receptor language understands the
result clearly. Finally, a translation is said to be natural when it uses the natural form, in terms of
grammar and style, of the receptor language which is easy to read.
© 2021 Cognizance Research Associates - Published by JLLS.709 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
1.3. Statement of the Problem
The gap that the present study intends to fulfill is to enrich and confirm the previous ones with
regard to the translation using GT from Indonesian into English and specifically focuses on academic
texts. Thus, the questions that arise are:
1. How does Google Translate render the Indonesian academic text into English?
2. To what extent does GT accurately transfer the academic texts from Indonesian into English?
3. To what extent does GT naturally transfer the academic texts from Indonesian into English?
2. Method
The study applies the descriptive – qualitative method. The source of data for this study is three
Indonesian academic articles written by three undergraduate lecturers from three disciplines:
Management, Japanese literature, and Mathematics. Then these articles are translated into English by
means of Google Translate web (https://translate.google.com). The data collected is limited to the
discussion section, since this part tends to be the writer's idea. In addition, only the textual information
is translated while all other details such as tables, pictures, and formulas are discarded. The data
analyzed for this paper is two sentences per article, thus there are six sentences to be analyzed. These
sentences are written in the discussion section and chosen randomly. After the English translation is
generated, the analysis is done by comparing the SL (Indonesian) and TL (English) with regard to
syntactic form and semantic meaning. In terms of form analysis, it is comparing the form in the SL and
that in the TL to find whether it is naturally built in the TL, while in terms of meaning, the messages in
the SL and TL are compared to find whether they are accurately transferred.
Larson's theory of translation assessment and evaluation (1998) is used as the reference for
accuracy and naturalness. The main theoretical framework used for analyzing the Indonesian sentence
structure is Sneddon, Adelaar, Djenar, & Ewing (2010) and Alwi, Lapoliwa, Dardjowidjojo, &
Moeliono (2003), while analyzing the English sentence structure uses Leech (2006).
3. Results and Discussion
There are all six sentences to be discussed. The following is the summary of the results regarding
three aspects: the translation types and strategy, the form, and the meaning.
Table 1. Result Summary
Data Type of sentence Kind/strategy Form Meaning Notes
(SL-TL) of translation
1A Simple - simple Literal Exactly the Fairly accurate;
same but inaccurate in one
natural noun
1B Complex - Literal Exactly the Fairly accurate;
Complex same but inaccurate in one
natural NP
2A Complex - Literal Exactly the Rather Incapable of
complex same, rather inaccurate, identifying topics of
unnatural especially in one discussion
NP
2B Compound - Literal Exactly the Inaccurate Most likely due to the
compound same, rather not well-formed SL,
unnatural thus it needs to be
© 2021 Cognizance Research Associates - Published by JLLS.Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719 710
paraphrased
3A Simple - simple Literal Not exactly the Accurate Replacement on one
same and VP and addition
natural
3B Complex - Literal Exactly the Fairly accurate; Changing the form of
complex same but inaccurate in one reduplication
natural verb
The data is analyzed per sentence as it is considered one unit of thought. Each sentence is firstly
categorized into sentence clause category which is based on the composition of clauses: simple,
compound, complex, and compound-complex.
Afterward, the discussion is divided into two kinds of analysis; the first one is a form or syntactic
analysis, and the second one is semantic or meaning analysis.
To facilitate the discussion, several abbreviations are used. They are:
GT: Google Translate
SL/ST: Source Language/Source Text
TL/TT : Target Language/Target Text
Syntactic Category:
N - NP : Noun - Noun Phrase
V - VP : Verb - Verb Phrase
A - AP : Adjective - Adjective Phrase
Adv - AdvP : Adverb - Adverb Phrase
P - PP : Preposition - Prepositional Phrase
Syntactic Function:
S : Subject
P : Predicate
O : Object
Adv : Adverbial
Data 1
Data 1 which contains two sentences (1A and 1B) is of a lecturer majoring in management.
Data 1A
Form and syntactic analysis
Indonesian : Dalam satu dekade terakhir, pertumbuhan pendapatan dan tren urbanisasi telah
mengubah preferensi konsumsi makanan penduduk Indonesia.
English : In the past decade, income growth and urbanization trends have changed the food
consumption preferences of the Indonesian population.
© 2021 Cognizance Research Associates - Published by JLLS.711 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
The type of sentence structure in the Indonesian text is simple since it has one subject
(pertumbuhan pendapatan dan tren urbanisasi) and one verb (telah mengubah). It is translated into a
simple sentence with 'income growth and urbanization trends' as subject and 'have changed' as a verb.
For this data, GT does not change the type of sentence.
Furthermore, the following table shows the syntactic category and function of Data 1A and its
translation.
Table 2. Syntactic category and function of Data 1A
Syntactic Syntactic Indonesian English
Category Function
PP Adverbial Dalam satu dekade terakhir In the past decade
NP Subject pertumbuhan pendapatan dan tren income growth and
urbanisasi urbanization trends
VP Predicator telah mengubah have changed
NP Direct Object preferensi konsumsi makanan penduduk the food consumption
Indonesia preferences of the Indonesian
population
In terms of syntactic category, the translation applies the exact category of the SL: PP-NP-VP-NP.
The syntactic function is also arranged in a similar way: Adverbial - Subject –Predicator – Direct
Object. It can be said that GT applies the same structure as that in the SL. Nevertheless, GT can adjust
the Indonesian NP structure into English. In Indonesian, the structure of an NP commonly puts the
head at the beginning which is followed by its modifier. It can be seen in the phrase pertumbuhan
pendapatan, with pertumbuhan as the head and pendapatan as its modifier. In English, the head noun
is generally placed at the end with its modifier being put before the head. The phrase is translated into
‘income growth’ with ‘growth’ as the head and ‘income’ as its modifier.
Furthermore, the Indonesian VP telah mengubah is translated into English VP 'have changed'.
Thus, GT in this case applies exactly the same structure.
In addition, the NP which functions as an Object, preferensi konsumsi makanan penduduk
Indonesia, is translated into an NP. However, GT has adjusted penduduk Indonesia into an English
structure using the preposition 'of' which shows possessive modifier, thus it is placed after the head
'preferences'.
Semantic Analysis
In terms of meaning, GT seems to be able to convey the original message of the SL although there
are some adjustments made. In the adverbial Dalam satu dekade terakhir, satu is not translated into
'one' but a definite article 'the'. This translation is semantically acceptable because in English 'the' is
used to specify the noun it refers to. Besides, tren in Indonesian is translated using plural form 'trends',
which in English is possible to use to show the general meaning of a noun. However, in the NP
preferensi konsumsi makanan penduduk Indonesia, penduduk is translated into 'population', which in
this context does not seem suitable. According to Oxford Dictionary 'population' is defined as 'all the
people who live in a particular area, city or country; the total number of people who live there.
(https://www.oxfordlearnersdictionaries.com/definition/english/population?q=population)
Based on this definition, population refers more to ‘number’ or ‘quantity’, while the message of the
original text is more about ‘people’. Therefore, the more acceptable translation of penduduk Indonesia
is ‘Indonesian people’.
© 2021 Cognizance Research Associates - Published by JLLS.Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719 712
Overall, in terms of form, the SL in data 1A can be said as naturally transferred even though GT
seems to do the literal translation, for the sequence of words and phrases is the same, except in the NP
the food consumption preferences of the Indonesian population. In terms of meaning, there is only one
word that is mistranslated (penduduk – population).
Data 1B
Form and syntactic analysis
Indonesian : Saat ini konsumsi daging, produk berbasis susu dan makanan siap saji yang termasuk
roti semakin lazim di Indonesia.
English : At present consumption of meat, dairy-based products, and prepared foods including
bread is increasingly prevalent in Indonesia.
The type of sentence 1B is complex since it contains two types of clauses: independent clause
konsumsi daging, produk berbasis susu dan makanan siap saji semakin lazim di Indonesia and
dependent clause yang termasuk roti which modifies NP makanan siap saji. The translation also
belongs to the complex sentence as it has an independent clause 'consumption of meat, dairy-based
products, and prepared foods is increasingly prevalent in Indonesia' and a dependent clause 'including
bread' which explains the NP 'prepared foods'.
The following table shows the syntactic category and function of Data 1B and its translation.
Table 3. Syntactic category and function of Data 1B
Syntactic Syntactic Indonesian English
Category Function
PP Adverbial Saat ini At present
NP Subject konsumsi daging, produk berbasis susu consumption of meat, dairy-
dan makanan siap saji yang termasuk based products, and prepared
roti foods including bread
AP Predicator semakin lazim is increasingly prevalent
PP Adverbial di Indonesia in Indonesia
In terms of syntactic category, the translation applies the exact category of the SL: PP-NP-AP-PP.
Likewise, the arrangement of the syntactic function of the TL is not different from that of the SL:
Adverbial – Subject – Predicator – Adverbial. Even the Indonesian NP konsumsi daging, produk
berbasis susu dan makanan siap saji yang termasuk roti is translated into a structure with the same
arrangement, using preposition ‘of’ that shows possessive modifier: ‘consumption of meat, dairy-based
products and prepared foods including bread’. In this phrase, however, there is a relative clause yang
termasuk roti which modifies NP makanan siap saji. GT applies the same structure with ‘including
bread’ as the relative clause explaining NP ‘prepared foods’. The only difference is that GT does not
include ‘which’ as the common translation of yang, thus the translated version applies a reduced
clause.
All in all, it can be said that in this complex sentence, GT applies the same arrangement as that in
the SL.
Semantic Analysis
In terms of meaning, the overall message seems to have been conveyed accurately. However, there
is an Indonesian NP makanan siap saji which GT translates into ‘prepared foods’. To some extent, this
© 2021 Cognizance Research Associates - Published by JLLS.713 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
translation is not accurate since makanan siap saji is commonly translated into ‘fast food’. As a matter
of fact, GT translates this phrase into ‘fast food’ when it is put on the web apart from the sentence and
paragraph. In addition, when ‘prepared foods’ is translated back into Indonesian, it becomes makanan
yang disiapkan, whose meaning is fairly different from makanan siap saji. Therefore, from the
semantic point of view, the message of the SL is not quite accurately transferred.
Overall, it can be said that in terms of form, the SL is naturally transferred even though GT uses the
same arrangement of the type of phrases. Only in the relative clause GT has reduced the clause. In
terms of meaning, GT does not totally transfer the message accurately due to the translation of
makanan siap saji into ‘prepared foods’.
Data 2
There are two sentences in this data (Data 2A and 2B). They are of a lecturer majoring in Japanese
Literature.
Data 2A
Form and syntactic analysis
Indonesian : Berdasarkan persepsi kesantunan tingkat pertama dengan konstruksi imperatif,
menduduki tingkat pertama paling santun adalah permintaan berpagar dan
permohonan secara eksplisit dengan jawaban responden sebanyak 96%.
English : Based on the perception of politeness of the first level with imperative construction,
occupying the first level of courtesy is a fenced request and an explicit request with a
respondent's answer of 96%.
The sentence of 2A belongs to Complex type because it consists of an independent clause,
menduduki tingkat pertama paling santun adalah permintaan berpagar dan permohonan secara
eksplisit and two dependent clauses: (1) Berdasarkan persepsi kesantunan tingkat pertama dengan
konstruksi imperatif which functions as adverbial of the sentence, and (2) dengan jawaban reponden
sebanyak 96% which functions as a relative clause. The English translation also has an independent
clause ‘occupying the first level of courtesy is a fenced request and an explicit request’. However,
there is only one dependent clause in the translation: ‘Based on the perception of politeness of the first
level with imperative construction’, while ‘with a response answer of 96%’ is not a clause, yet a PP
describing the noun ‘a fenced request and an explicit request’.
The following table shows the syntactic category and function of Data 2A and its translation.
Table 4. Syntactic category and function of Data 2A
Syntactic Syntactic Indonesian English
Category Function
Adv. Clause Adverbial Berdasarkan persepsi kesantunan Based on the perception of politeness of the
tingkat pertama dengan first level with imperative construction
konstruksi imperatif
NP Subject menduduki tingkat pertama occupying the first level of courtesy
paling santun
Copula Predicator adalah is
NP Subject permintaan berpagar dan a fenced request and an explicit request
Complement permohonan secara eksplisit
© 2021 Cognizance Research Associates - Published by JLLS.Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719 714
PP Noun dengan jawaban responden with a respondent's answer of 96%.
Complement sebanyak 96%.
In terms of syntactic category, GT has arranged the sentence into Adv.-NP-Copula-NP-PP.
Basically, this arrangement is the same as that in the SL. Interestingly, the subject in the SL is
occupied by a verb menduduki which is translated into gerund 'occupying'. In this case, it is obvious
that GT literally translates the verb which functions as NP into the same category. In another case,
sebanyak 96% which occurs in the PP functioning as Noun Complement, is categorized as a classifier
in Indonesian (Sneddon et al., 2010, p.145), yet it is translated as ‘of 96%’ which becomes a part of the
NP ‘a respondent’s answer’.
Semantic Analysis
The problem of the meaning translated by GT is observed on the SL’s NP tingkat Pertama, both in
the adverbial and in the subject. GT has translated this phrase into 'the first level', which is not quite
accurate since the sentence is written in a context discussing the result of a survey that has generated a
number showing a certain position or rank in politeness strategies. It can be said that GT is not yet
capable to identify the context being discussed in the text. In addition, paling santun which is a part of
the NP functioning as Subject is translated into 'courtesy'. Even though 'courtesy' is one of the
synonyms of 'politeness', it does not match the context for politeness is a specific term in
sociolinguistic study. Moreover, it seems that this translation is due to the SL structure which is not
well-formed. As a result, GT translates as it is, which shows that GT follows a literal translation
procedure. As the topics discussed in the SL deals with politeness, what is meant by the NP is most
likely of menduduki tingkat pertama kesantunan or menduduki tingkat paling santun. Therefore, the
translation of the phrase should be: ‘occupying the first rank of politeness’ or ‘occupying the
rank/position of being the most polite’.
In another phrase, NP functioning as SC, permintaan berpagar is translated into ‘a fenced request’.
Again, GT has shown that it follows the literal translation procedure – Pagar in Indonesian means
‘fence’ in English. However, since the context of the text deals with politeness theory, ‘a fenced
request’ becomes inaccurate. In politeness theory, there is a term ‘hedges’ which is one of some
politeness strategies. It refers to mitigating devices to soften the force of a statement, or as boosters
when the function is emphatic. (Holmes, 2013) Thus, permintaan berpagar should be translated into
‘hedged request’ or simply ‘indirect request’.
Data 2B
Form and syntactic analysis
Indonesian : Selanjutnya menggunakan isyarat kuat 70% responden menjawab, menggunakan
modus imperatif 50% responden, dan menggunakan pernyataan keinginan implisit
sebanyak 43% responden.
English : Furthermore, using a strong signal 70% of respondents answered, using the
imperative mode of 50% of respondents, and using an implicit desire statement of
43% of respondents.
Sentence 2B can be categorized into compound sentence because it consists of three independent
clauses: (1) menggunakan isyarat kuat 70% responden menjawab, (2) menggunakan modus imperatif
50% responden, and (3) menggunakan pernyataan keinginan implisit sebanyak 43 % responden.
These sentences are connected using the conjunction dan. The translation also contains three
independent clauses: (1a) ‘using a strong signal 70% of respondents answered’, (2a) ‘using the
© 2021 Cognizance Research Associates - Published by JLLS.715 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
imperative mode 50% of respondents’, and (3a) ‘using an implicit desire statement of 43% of
respondents’. These are connected by the conjunction 'and. Obviously, GT applies the same type of
sentences.
The following table shows the syntactic category and function of Data 2B and its translation.
Table 5. Syntactic category and function of Data 2B
Syntactic Syntactic Indonesian English
Category Function
Conjunction Connector Selanjutnya Furthermore
NP Subject1 menggunakan isyarat kuat using a strong signal
NP Predicate1 70% responden menjawab 70% of respondents answered
NP Subject2 menggunakan modus imperatif using the imperative mode
NP Predicate2 50% responden 50% of respondents
Conjunction Connector dan and
NP Subject3 menggunakan pernyataan keinginan using an implicit desire
implisit statement
Classifier/PP Predicator3 sebanyak 43% responden of 43% of respondents
From the table above, it is indicated that GT has translated the SL using the same form. The
category and function of each phrase in the SL and in the TL are almost the same. Menggunakan, for
example, is actually a verb that functions similarly as a Noun (Subject) and it is translated literally into
'using' whose form is a gerund – a verb functioning as a noun. However, in the last phrase sebanyak
which is categorized as a classifier by Sneddon et al. (2010, p.145) is translated into ‘of’ which
belongs to PP.
Semantic Analysis
The problem is the meaning conveyed in the TL can be seen in the first NP: '70% of respondents
answered’. GT seems to apply the literal translation since it translates the Indonesian verb menjawab
into 'answered' which in English is understood as a verb also. On the other hand, this may be due to the
SL sentence which is not quite well-formed. Based on Indonesian grammar (Alwi et al., 2003) the first
independent clause can be paraphrased into: Ada 70% responden yang menjawab menggunakan
isyarat kuat. GT has translated it into ‘There are 70% of respondents who answered using strong cues'.
(translated on 20 November 2019) This version of translation seems to be more accurate compared to
the previous one. However, the difference lies in the difference in the SL's structure.
Data 3
Data 3 consists of two sentences (3A and 3B) collected from a lecturer majoring in mathematics.
Data 3A
Form and syntactic analysis
Indonesian : Pada subbab ini akan diberikan beberapa contoh kasus untuk disimulasikan.
English : In this section, there will be some case examples to be simulated.
© 2021 Cognizance Research Associates - Published by JLLS.Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719 716
The sentence in 3A belongs to a simple sentence as it has one independent clause akan diberikan
beberapa contoh kasus. The translation also has one independent clause ‘there will be some case
examples’. Thus, in this sentence GT exactly follows the type of the SL sentence.
The following table shows the syntactic category and function of data 3A and its translation.
Table 6. Syntactic category and function of Data 3A
Syntactic Syntactic Function Indonesian English
Category
PP Adverbial Pada subbab ini In this section
Existential Dummy S - there
VP Predicator akan diberikan will be
NP Inverted/Extraposed beberapa contoh kasus some case examples
Subject
VP Noun complement untuk disimulasikan to be simulated
For this translation, GT uses almost the same arrangement. The only difference is on the occurrence
of 'there' which in English is called existential there which functions as a dummy subject. (Leech,
2006, p. 39-40) The real subject is beberapa contoh kasus translated as ‘some case examples’. In the
SL beberapa, contoh kasus is placed after the verb akan diberikan, hence there is an inversion between
Predicator and Sentence Subject. In the TL, ‘some case examples’ is placed after the verb ‘will be’,
which follows the same arrangement.
Semantic analysis
The message of the SL in this sentence seems to have been accurately translated by GT. There is
only a change in the translation which occurs in the verb 'will be as the translation of akan diberikan.
GT here has omitted the verb diberikan and replaces it with the linking verb 'be' and also added
existential 'there' in front of the verb. Since the meaning in the TL is acceptable and accurate, this kind
of replacement is one procedure that GT has applied.
Data 3B
Form and syntactic analysis
Indonesian : Simulasi ini dilakukan untuk mengetahui kecenderungan perubahan arus lalu lintas
jika data diubah - ubah.
English : This simulation is carried out to determine the trend of changes in traffic if the data is
changed.
The sentence is categorized into complex one since it has one independent clause Simulasi ini
dilakukan untuk mengetahui kecenderungan perubahan arus lalu lintas and one dependent clause jika
data diubah – ubah. The translation also follows the same structure, with ‘This simulation is carried
out to determine the trend of changes in traffic’ as the independent clause, and ‘if the data is changed’
as the dependent one.
The following table shows the syntactic category and function of data 3B and its translation.
© 2021 Cognizance Research Associates - Published by JLLS.717 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
Table 7. Syntactic category and function of Data 3B
Syntactic Syntactic Function Indonesian English
Category
NP Subject Simulasi ini This simulation
VP passive predicator dilakukan is carried out
VP Purpose clause untuk mengetahui kecenderungan to determine the trend of
perubahan arus lalu lintas changes in traffic
Dependent Conditional clause jika data diubah - ubah if the data is changed.
clause
The structure in the TL seems to follow that in the SL: NP – VP – VP (purpose) – Clause. Once
again, GT applies the literal translation since it uses the same form.
Semantic Analysis
The message in the TL seems to be acceptable and understood as that in the SL although the form
applied by GT is exactly the same. Yet, there is one message which will be differently understood.
diubah-Ubah in Indonesian may mean ‘repeatedly done’ (Alwi et al., 2003; Sneddon et al., 2010), so
the message of the SL means that the data can be changed more than once. The author of the text may
apply some different data to the formula simulation. However, GT has translated it into 'changed',
which may indicate that the changes only happen once. Thus, the translation should become: 'if the
data is changed repeatedly', or more freely: 'if different data is inputted into the formula'. It can be said
that GT is not sensitive toward the different meanings of Indonesian diubah and diubah-ubah.
4. Conclusions
The findings generally indicate that Google Translate applies literal translation. It can be seen from
the overall translations of sentences which literally adapt the forms of the Source Language. Only a
few phrases use different forms. Even though there is a strategy of replacement and addition, GT has
translated mostly literally. In terms of meaning, the overall messages seem to be fairly acceptable
because there are a few words that are translated inaccurately. From these findings, it can be said that
GT still needs improvement, especially concerning forms and context-based meaning because GT is
not yet capable of identifying the context or topics being discussed in the text. One thing to be
considered carefully is that the structure in the SL needs to be reviewed first before the source text is
inputted in GT because this machine is still translated literally. Thus, the text in the SL must be well-
formed to achieve a better translation.
To some extent the findings of this study support and confirm the findings of (1) Khosravizadeh &
Pashmforoosh (2011) in that the translation should consider the TL’s precise structure of syntactic
units before dealing with semantics problems. It should also be able to convey the context-embedded
lexical expression; (2) Vidhayasai et al. (2015) with regard to the inaccuracy that occurs in lexical
level and unnaturalness in syntactic level; (3) Groves & Mundt (2015) implying that GT is not yet able
to translate by aligning the norms at discourse level; (4) Rensburg et al. (2012) in that the result of
machine translation requires post-editing by human or professional translators and inputs from clients.
In addition, the findings also confirm that of Li et al. (2014) in that raw results of GT can be used to
convey general information only if the accuracy of grammatical aspects is less taken into account.
© 2021 Cognizance Research Associates - Published by JLLS.Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719 718 5. Pedagogical Implication Nowadays, it seems that Google Translate or other Machine Translation has become a part of the translation process done by beginners, even experts in the translation field often use the machine. The present study deals with the Indonesian lecturer's academic writing. The result can be a valuable input to academic institutions or language trainers who deal with lecturer language training, especially in writing academic English. GT can be used as a tool to translate Indonesian texts into English. However, the lecturers, as well as the trainers, are not suggested to fully rely on GT. The translation generated by GT needs a process of reviewing and editing which can be done by professional translators or language experts. References Allue, B. R. (2017). The Reliability and Limitations of Google Translate: A Bilingual, Bidirectional and Genre-Based Evaluation. Entreculturas, 9(February), 67–80. Alwi, H., Lapoliwa, H., Dardjowidjojo, S., & Moeliono, A. M. (2003). Tata Bahasa Baku Bahasa Indonesia (3rd ed.). Balai Pustaka. Arsyad, S., Purwo, B. K., Sukamto, K. E., & Adnan, Z. (2019). Factors hindering Indonesian lecturers from publishing articles in reputable international journals. Journal on English as a Foreign Language, 9(1), 42–70. Awadh, A. N. M., & Shafiull, K. A. (2020). Challenges of translating neologisms comparative study: Human and machine translation. Journal of Language and Linguistic Studies, 16(4), 1987–2002. Calude, A. S. (2003). Machine Translation of Various Text Genres. 7th Language and Society Conference of the New Zealand Linguistic Society, January. Chen, X., Acosta, S., & Barry, A. E. (2016). Evaluating the Accuracy of Google Translate for Diabetes Education Material. JMIR Diabetes, 1(1), 1–11. Erton, İ., & Tanbi, Y. (2016). Significance of Linguistics in Translation Education at the University Level. Journal of Language and Linguistic Studies, 12(2), 38–53. Groves, M., & Mundt, K. (2015). Friend or foe? Google Translate in language for academic purposes. English for Specific Purposes, 37, 112–121. Hartono, Arjanggi, R., & Praptawati, D. (2019). Self-Efficacy of Indonesian Non-English Lecturers in Writing English Academic Papers for International Publication. Advances in Social Science, Education and Humanities Research, 188(ELTLT 2018), 28–31. Holmes, J. (2013). An Introduction to Sociolinguistics (4th ed.). Routledge. Hutchins, W. J. (2006). Machine Translation: History. In Encyclopedia of Language & Linguistics, 7, (2nd ed., pp. 375–383). Oxford: Elsevier. Khosravizadeh, P., & Pashmforoosh, R. (2011). Google translation : A semantic structure analysis. Lingoistica, July. Kuczmarski, J. (2018). A new look for Google Translate on the web. https://www.blog.google/products/translate/new-look-google-translate-web Leech, G. (2006). A Glossary of English Grammar. Edinburgh University Press. Li, H., Graesser, A. C., & Cai, Z. (2014). Comparison of Google Translation with Human Translation. © 2021 Cognizance Research Associates - Published by JLLS.
719 Winiharti et al. / Journal of Language and Linguistic Studies, 17(2) (2021) 706-719
Proceedings of the Twenty-Seventh International Florida Artificial Intelligence Research Society
Conference, 190–195.
Lotz, S., & Rensburg, A. van. (2014). Translation technology explored : Has a three-year maturation
period done Google Translate any good ? Linguistics Plus, 43, 235–259.
Panter, M. (n.d.). Translating Academic Writing: Human vs. Machine. Retrieved November 18, 2019,
from https://www.aje.com/arc/translating-academic-writing-human-vs-machine/
Rensburg, A. Van, Snyman, C., & Lotz, S. (2012). Applying Google Translate in a higher education
environment : Translation products assessed. Southern African Linguistics and Applied Language
Studies, 30(4), 511–524.
Sneddon, J. N., Adelaar, A., Djenar, D. N., & Ewing, M. C. (2010). Indonesian Reference Grammar
(2nd ed.). Allen & Unwin.
Vidhayasai, T., Keyuravong, S., & Bunsom, T. (2015). Investigating the Use of Google Translate in
“Terms and Conditions” in an Airline’s Official Website: Errors and Implications. 49(June),
PASAA Journal of Language Teaching and Learning.
Voigt, R., & Jurafsky, D. (2012). Towards a Literary Machine Translation : The Role of Referential
Cohesion. CLfL@NAACL-HLT.
AUTHOR BIODATA
Menik Winiharti is a lecturer at English Department, Bina Nusantara University, Jakarta. She is now pursuing
her doctoral degree in linguistics at Indonesia University of Education, Bandung. Her research interests are
linguistics in general and English language skills, as well as English language learning.
Syihabuddin is a professor of translation from the Indonesian Department, Indonesia University of Education,
Bandung. His prominent work is translating Tafsir Ibnu Katsir from Arabic into Indonesian.
Dadang Sudana is a senior lecturer at the Linguistic Department, Indonesia University of Education, Bandung.
His research interests are linguistics in general and Language acquisition.
© 2021 Cognizance Research Associates - Published by JLLS.You can also read
